A Comprehensive Evaluation Method for the Similarity of a Bionic Vehicle from Multiple Perspectives

Abstract

The present invention relates to a comprehensive evaluation method for the similarity of a bionic vehicle from multiple perspectives. First, images are collected from different perspectives of the bionic vehicle and real organisms, their contours are extracted, and as uniform contour sampling points as possible are obtained through the contour point sampling method. Secondly, the shape context algorithm is used to calculate the similarity evaluation matrix. Finally, the similarity between each view is calculated through the shape context distance. The similarities of each view are weighted and summed to obtain the overall shape similarity of the bionic vehicle. This method calculates the similarity value of the bionic vehicle based on the shape context algorithm, indicating the direction for the optimization of its shape and structure.

Description

Technical Field

[0001] The present invention belongs to the similarity evaluation method of underwater vehicles, and relates to a comprehensive evaluation method for the similarity of bionic vehicles from multiple perspectives. Background Art

[0002] As an intelligent underwater vehicle platform, unmanned underwater vehicles have been widely used in military and civilian operation tasks such as resource exploration and enemy reconnaissance, such as hydrographic information collection, underwater and surface target monitoring, etc. In order to further improve the efficiency, mobility and biocompatibility of the vehicle, bionic underwater vehicles have become an international research hotspot.

[0003] With the in-depth research of bionic vehicles, the improvement of similarity can greatly improve their stability and flexibility. Establishing a comprehensive evaluation method for the similarity of bionic vehicles from multiple perspectives can provide a basis for formulating the shape design method of bionic vehicles. The traditional bionic similarity evaluation method has uneven sampling of contour points and cannot accurately evaluate the similarity. In order to measure the bionic similarity, the traditional method generally extracts feature points in the contour through information such as curvature for similarity evaluation. However, this method only uses specific information of a certain part of the contour and does not consider the global information of the entire graph, resulting in the bionic similarity evaluation result being difficult to accurately reflect the similarity and difference between the bionic vehicle and the biological prototype. Therefore, using the traditional method for bionic similarity evaluation is not accurate enough, and a new similarity evaluation method needs to be designed to provide a scientific decision-making basis for the research and design of bionic vehicles. In the published literature, there is no example of achieving comprehensive similarity evaluation by obtaining global information through multi-perspective observation. Summary of the Invention

[0004] Technical Problems to be Solved

[0005] In order to avoid the deficiencies of the prior art, the present invention proposes a comprehensive evaluation method for the similarity of bionic vehicles from multiple perspectives to achieve the bionic similarity evaluation of bionic vehicles and real organisms.

[0006] The characteristics of the method are as follows: In order to ensure uniform sampling of the contour, the Jitendra sampling method is used to sample the contour points. In order to use the global information of the contour map for similarity evaluation, the shape context algorithm is used to calculate the shape context distance, and combined with the similarity values corresponding to the three views, the overall similarity value of the bionic vehicle is obtained.

[0007] Technical Solution

[0008] A comprehensive evaluation method for the similarity of bionic vehicles from multiple perspectives, characterized by the following steps:

[0009] Step 1: Respectively collect the three views of the three-dimensional appearance of the real organism and its bionic vehicle. Apply the Canny operator to each view to obtain the contours of each view of the bionic vehicle and the real organism. Subsequently, use the Jitendra sampling method to sample the contour points of each view to obtain the contour point sampling maps of their respective three views;

[0010] Step 2: Construct a similarity evaluation matrix: For each contour point sampling map, with each contour point as the center, construct a logarithmic polar coordinate system, and divide the logarithm of the distance log r between this point and the remaining sampling points into multiple regions;

[0011] Starting from directly above the center point, divide 360° into multiple regions in the clockwise direction, and then map the remaining contour points to each region;

[0012] Count the number of contour points falling in each region to construct a shape context histogram;

[0013] Divide the number of contour points in each region by the number of contour points falling in all regions for normalization to generate the shape context matrix under this view;

[0014] Implement Step 2 for all contour point sampling maps to obtain the shape context matrices of the bionic vehicle and the organism under three views.

[0015] Step 3: For each point pair in the contour image of the bionic vehicle, correspond a histogram vector, denoted as g i (k); For each point pair in the contour image of the real organism, correspond a histogram vector, denoted as h i (k); Use the chi-square formula to calculate the similarity evaluation matrix to obtain the similarity evaluation matrix C of the shape context under different views s :

[0016]

[0017] Where: g i and h i represent the contour points of the bionic vehicle and the real organism under different views, and i represents different views;

[0018] Step 4: Calculate the similarity of a single view of the bionic vehicle: Calculate the normalized shortest distance Dist between each point pair of the bionic vehicle and the real organism min :

[0019]

[0020] Step 5: The similarity metric matrix C sThe sum of the minimum loss values for each row and the sum of the minimum loss values for each column are added together and then divided by the number of sampling contour points N to obtain the shape context distance D sc It is obtained through the following formula:

[0021]

[0022] where: G i and H i respectively represent the contours collected from different views of the bionic vehicle and the real creature;

[0023] Step 6: Calculate the similarity constructed by the shape context algorithm:

[0024]

[0025] Step 7: Design the similarity of the external shape features between the real creature and the bionic vehicle as the weighted sum of the similarity values of the three views, assign different weight values to each view, and calculate the overall similarity value of the bionic vehicle:

[0026] S S = aS0 + bS1 + cS2

[0027] where S s represents the overall similarity, S0 represents the similarity of the top view, S1 represents the similarity of the side view, S2 represents the similarity of the front view, and a, b, and c are the weight values of the top view, side view, and front view respectively.

[0028] The logarithmic value log r in step 2 is divided into multiple regions, specifically 5 regions.

[0029] The division of 360° into multiple regions in step 2 is as follows: each 30° is divided into a region, the angle value θ is divided into 12 regions, and then the remaining contour points are mapped to each region.

[0030] The weight value of the top view is greater than the weight values of the side view and the front view.

[0031] The weight values of the top view, side view, and front view are: a = 0.6, b = 0.2, c = 0.2.

[0032] Beneficial effects

[0033] A comprehensive evaluation method for the similarity of a bionic vehicle from multiple perspectives proposed by the present invention first collects images from different perspectives of the bionic vehicle and real organisms, extracts their contours, and obtains as uniform contour sampling points as possible through the contour point sampling method. Secondly, the shape context algorithm is used to calculate the similarity evaluation matrix. Finally, the similarity between each view is calculated through the shape context distance. The similarities of each view are weighted and summed to obtain the overall shape similarity of the bionic vehicle. This method calculates the similarity value of the bionic vehicle based on the shape context algorithm, indicating the direction for the optimization of its shape and structure.

[0034] Compared with the prior art, the present invention has the following beneficial technical effects:

[0035] 1. Using the Jitendra sampling method to sample the contour points can ensure that the sampling points are uniform enough, avoiding the problem that the contour sampling points may be too concentrated due to random sampling. The obtained sampling points can fully reflect the characteristic information contained in the entire contour;

[0036] 2. Constructing the shape context operator and using the global information of the contour graph rather than the information of a few feature points for similarity evaluation makes the evaluation result more accurate;

[0037] 3. Using the point-to-point distance for matching, the obtained normalized shortest distance can reflect the difference degree between the bionic vehicle and the real organism, facilitating the quantitative analysis of the parts with large shape differences between the two;

[0038] 4. Through the comprehensive evaluation of similarity from multiple perspectives, the shape features of the bionic vehicle and real organisms can be fully collected, and the evaluation result is more scientific and reasonable. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is the flowchart of the similarity evaluation of the bionic vehicle based on the shape context algorithm of the present invention.

[0040] Figure 2 is the side view contour point sampling diagram of the manta ray-like vehicle.

[0041] Figure 3 is the logarithmic polar coordinate system obtained with a point in the contour sampling points of the manta ray-like vehicle as the center.

[0042] Figure 4 is the shape context histogram obtained with a point in the contour sampling points of the manta ray-like vehicle as the center.

[0043] Figure 5 is the front view of the manta ray-like vehicle.

[0044] Figure 6 is the side view of the manta ray-like vehicle.

[0045] Figure 7 It is a top view of a manta ray - inspired vehicle.

[0046] Figure 8 It is a schematic diagram of the difference between the front - view contour of the manta ray - inspired vehicle and the front - view contour of the manta ray.

[0047] Figure 9 It is a schematic diagram of the difference between the side - view contour of the manta ray - inspired vehicle and the side - view contour of the manta ray.

[0048] Figure 10 It is a schematic diagram of the difference between the top - view contour of the manta ray - inspired vehicle and the top - view contour of the manta ray. Detailed implementation manners

[0049] The present invention will be further described in combination with embodiments and drawings:

[0050] The technical solution adopted by the present invention is to collect the three - view drawings of the bionic vehicle and the real organism, convert the three - dimensional appearance into two - dimensional views; perform contour extraction and contour point sampling to obtain relatively uniform contour sampling results of the bionic vehicle and the real organism; construct a shape context operator to obtain a shape context matrix for calculating the shape context distance; use the shape context distance to calculate the bionic similarity of each view of the bionic vehicle, and then calculate the overall similarity value.

[0051] To comprehensively and clearly present the purpose, technical solution and advantages of the present invention, the following will further elaborate on the specific implementation manners of the multi - perspective similarity comprehensive evaluation method for bionic vehicles in combination with the drawings. It should be noted in advance that for the convenience of description, the given drawings are only partial structural schematic diagrams related to the present invention, not all embodiments of the present invention. The specific steps are as follows:

[0052] Refer to Figure 1 , a multi - perspective similarity comprehensive evaluation method for bionic vehicles, including the following steps:

[0053] 1: Collect the three - view drawings of the bionic vehicle and the real organism, convert their three - dimensional appearance into two - dimensional views, and use the Canny operator to obtain the three - view contours of the bionic vehicle and the real organism. Subsequently, use the Jitendra sampling method to perform contour point sampling on the three - view contours

[0054] Collect the two - dimensional views of the bionic vehicle and the real organism from the top - view, front - view and side - view Figure 3 angles, and convert the three - dimensional model into two - dimensional views.

[0055] Use the Canny operator to obtain the three - view contours of the bionic vehicle and real organisms. Subsequently, use the Jitendra sampling method to sample 100 contour points from the three - view contours. Figure 2 Figure 5 shows the sampling diagram of the side - view contour points of the manta - ray - like vehicle.

[0056] 2. Construct a similarity evaluation matrix. For each contour point, construct a log - polar coordinate system centered on it. Count the number of contour points falling into each region to construct the shape context operator. Finally, divide by the number of contour points falling into all regions for normalization to generate the shape context matrix.

[0057] Specifically, for each contour point of a certain view of the bionic vehicle and real organisms, construct a log - polar coordinate system centered on it. Divide the logarithm of the distance log r between this point and the remaining sampling points into 5 regions. Starting from directly above the center point and moving clockwise, divide each 30° as a region. Divide the angular value θ into 12 regions, and then map the remaining contour points into each region. Figure 3 Figure 6 shows the log - polar coordinate system obtained with a point in the contour sampling points of the manta - ray - like vehicle as the center. Figure 4 Figure 7 shows the shape context histogram obtained with a point in the contour sampling points of the manta - ray - like vehicle as the center.

[0058] Furthermore, finally divide the number of contour points in each region by the number of contour points falling into all regions for normalization to generate the shape context matrix for this view.

[0059] Perform the above operations on each contour point of the three - view diagrams to obtain the shape context matrices of the bionic vehicle and organisms under the three views.

[0060] Furthermore, for each pair of points in the contour images of the bionic vehicle and organisms, there is a histogram vector, denoted as g i (k) and h i (k) respectively. Substitute them into the chi - square formula to calculate the similarity metric matrix, and a similarity evaluation matrix C regarding the shape context can be obtained s :

[0061]

[0062] where g i and h i represent the contour points of the bionic vehicle and real organisms under different views.

[0063] 3. Define the normalized shortest distance Dist between each pair of points of the bionic vehicle and real organisms min as follows:

[0064]

[0065] For the similarity metric matrix C s , add the sum of the minimum loss values of each row and the sum of the minimum loss values of each column, and then divide by the number of sampling contour points N. The shape context distance D sc is obtained by the following formula:

[0066]

[0067] where G i and H i represent the contours collected from different views of the biomimetic vehicle and the real organism respectively.

[0068] The similarity calculation method constructed by the shape context algorithm is as follows:

[0069]

[0070] 4. Design the similarity of the external shape features between the real organism and the biomimetic vehicle as the weighted sum of the similarity values of three views. Figure 5 、 Figure 6 、 Figure 7 show the front view, side view, and top view of the collected manta ray biomimetic vehicle.

[0071] By assigning different weighting values to each view, calculate the overall similarity value of the biomimetic vehicle as follows:

[0072] S S = aS0 + bS1 + cS2

[0073] where S s represents the overall similarity, S0 represents the top view similarity, S1 represents the side view similarity, S2 represents the front view similarity, and a = 0.6, b = 0.2, c = 0.2 are the weights of each view. Table 1 in the appendix gives the similarity values of each view and the overall similarity value of the manta ray biomimetic vehicle in this embodiment.

[0074] Combining the Dist min values of each view of the biomimetic vehicle, the parts with large differences between the 3D model of the biomimetic vehicle and the real organism can be obtained. Figure 8 、 Figure 9 、 Figure 10 show the differences between the three views of the manta ray biomimetic vehicle and the three views of the real manta ray.

[0075] The specific embodiments described above are only for showing and explaining the technical idea of the present invention, and cannot be used to limit the present invention. Any modifications, substitutions, and improvements made to the technical solution within the design idea and principle of the present invention shall fall within the protection scope of the present invention.

[0076] Table 1 Similarity values of the manta ray biomimetic vehicle

[0077]

Claims

1. A comprehensive evaluation method for multi - perspective similarity of bionic vehicles, characterized in that The steps are as follows: Step 1: Respectively collect the three views of the three-dimensional appearance of the real organism and its bionic vehicle. Apply the Canny operator to each view to obtain the outlines of each view of the bionic vehicle and the real organism. Subsequently, use the Jitendra sampling method to sample the contour points of each view of the outline, and obtain the contour point sampling maps of their respective three views; Step 2: Construct a similarity evaluation matrix: For each contour point sampling map, with each contour point as the center, construct a log-polar coordinate system, and divide the logarithm of the distance log r between this point and the remaining sampling points into multiple regions; Starting from directly above the center point, divide 360° into multiple regions in the clockwise direction, and then map the remaining contour points to each region; Count the number of contour points falling in each region to construct a shape context histogram; Divide the number of contour points in each region by the number of contour points falling in all regions for normalization to generate the shape context matrix under this view; Implement Step 2 for all contour point sampling maps to obtain the shape context matrices of the bionic vehicle and the organism under the three views; Step 3: Corresponding each point pair in the contour image of the bionic vehicle to a histogram vector, denoted as g i (k); Corresponding each point pair in the contour image of the real creature to a histogram vector, denoted as h i (k); Using the chi-square formula to calculate the similarity evaluation matrix, and obtaining the similarity evaluation matrix C of the shape context under different views s : where: g i and h i represent the contour points of the bionic vehicle and the real creature in different views, and i represents different views; Step 4, calculate the single-view similarity of the bionic vehicle: Calculate the normalized shortest distance Dist between each point pair of the bionic vehicle and the real creature min : Step 5: Take the similarity metric matrix C s Add the sum of the minimum loss values for each row and the sum of the minimum loss values for each column, and then divide by the number of sampling contour points N to obtain the shape context distance D sc which is obtained through the following formula: Among them: G i and H i respectively represent the outlines collected from different views of the bionic vehicle and the real organism; Step 6: Calculate the similarity constructed by the shape context algorithm: Step 7: Design the similarity of the external shape features between the real organism and the bionic vehicle as the weighted sum of the similarity values of the three views. Assign different weighted values to each view and calculate the overall similarity value of the bionic vehicle: S S = aS0 + bS1 + cS2 Among which S s represents the overall similarity, S0 represents the similarity of the top view, S1 represents the similarity of the side view, S2 represents the similarity of the front view, and a, b, and c are the weights of the top view, side view, and front view respectively.

2. The comprehensive evaluation method for the similarity of a bionic vehicle from multiple perspectives according to claim 1, characterized in that: The logarithm log r in Step 2 is divided into 5 regions.

3. The comprehensive evaluation method for the multi-view similarity of the bionic vehicle according to claim 1, wherein: The division of 360° into multiple regions in Step 2 is: Divide each 30° into a region, divide the angle value θ into 12 regions, and then map the remaining contour points to each region.

4. The comprehensive evaluation method for the multi - perspective similarity of the bionic vehicle according to claim 1, wherein: The weight of the top view is greater than the weights of the side view and the front view.

5. The comprehensive evaluation method for the multi-perspective similarity of the bionic vehicle according to claim 1 or 4, characterized in that: The weights of the top view, side view, and front view are: a = 0.6, b = 0.2, c = 0.2.

Images